Radical Functions
Learning goals
- Solve radicand for an even-index domain
- Accept every real input for an odd-index root
- Draw from its corner, rising ever slower
- Place the corner at in
- Check every candidate, since squaring adds extraneous ones
What a radical function is
Squaring sends to and to . The square root asks the reverse question: which nonnegative number, squared, gives the input? Packaging that question as a function gives the square-root function
By agreement means the principal root, the one that is not negative. Both and square to , but is defined to be alone. That definition makes return exactly one output per input, so is a genuine function.
The same idea works for any power. The cube root undoes cubing, and in general the th root undoes raising to the th power. The little number is the index of the radical; a square root is the case , and its index is left unwritten. Each of these is the same thing you already know from rational exponents, since taking an th root is raising to the power :
A radical function is any function whose rule contains the variable under a root like this. The rest of the lesson works out the two features that make radicals behave unlike the polynomials of earlier chapters: a restricted domain, and a graph that bends.
Domain: which inputs a root will accept
A polynomial such as accepts every real number. A radical is fussier. Try to compute and you are asking for a number whose square is , and no real number squares to a negative. So has no real value, and cannot be in the domain of . The same objection rules out every negative input.
Why an even-index root rejects negative inputs#
Start from the definition. The value is the nonnegative number whose square is , so if then must satisfy . The question is whether any real number can have a negative square. A positive times a positive is positive, a negative times a negative is again positive, and , so for every real number ,
A square is never negative. If is negative, there is no real with , so has no real value. That is exactly why the domain of is every with . The same reasoning applies to any even index: the input under a fourth root, a sixth root, and so on must be zero or greater.
Odd-index roots escape the restriction, and the sign test shows why. A cube can be negative, since , which makes a perfectly good real number. As runs through every real number, runs through every real output exactly once, so is defined for all real . The domain shrinks only when the index is even, because only an even power is forced to stay nonnegative.
To find the domain of a more involved radical, you do not need a new rule. For an even root, whatever sits under the radical must be greater than or equal to zero. So for an even root, set the radicand (the expression under the root) to be nonnegative and solve the resulting inequality.
Worked example 1 Find the domain of three radical functions
Handle each root by looking only at what it demands of the radicand.
For , the index is even, so the radicand cannot be negative. Require it to be nonnegative and solve:
The domain is all .
For , again set the radicand to be nonnegative, and remember that dividing by a negative flips the inequality:
The domain is all . Notice the domain runs to the left here, because raising makes the radicand smaller.
For , the index is odd, so the cube root accepts every real number no matter the sign of the radicand. The domain is all real numbers, with nothing to exclude.
Check your understanding
What is the domain of ?
The index is even, so the radicand must be zero or greater. Set it nonnegative and solve.
Dividing by the positive keeps the inequality direction, so the domain is every .
The graph of the square root
To plot , choose inputs whose roots come out whole, the perfect squares, so the points are easy to place:
The curve begins at the corner , since , and it climbs to the right forever. But the climb keeps slowing: getting from output to output costs three units of input (from to ). Getting from output to output then costs five units of input (from to ). Each additional unit of height demands a wider stretch of input, so the graph rises steeply near the start and then flattens.
That shape is not arbitrary. Saying says two things at once: (the root is never negative) and (squaring the output returns the input). Read on its own, the equation describes a parabola lying on its side, opening to the right, because for each height the input is . Keeping only the part with leaves the upper half of that sideways parabola, and that half is exactly the square-root graph.
There is a second way to see the same curve. Since the square root undoes squaring on nonnegative inputs, is the inverse of restricted to . So the graph of is the reflection of that parabola across the line , the mirror you met when studying inverse functions. Reflecting swaps each point into , which turns the steep, upward half-parabola into the flat, rightward square-root curve.
The graph also settles the range. The outputs of are the heights the curve reaches, and starting at it climbs through every positive value without bound, so the range is every . Every nonnegative number is hit exactly once, since a target height is the root of .
Reading transformations off the equation
You do not have to build a fresh table for every radical function. Each one is a shifted and scaled copy of . The moves behind that copy are the same shifts, stretches, and reflections from the graphing lessons, now applied to the root. The general form is
Read the three pieces separately, exactly as you did for parabolas:
- The inside the root shifts the graph right by . It also sets the domain, because the radicand must be nonnegative, giving .
- The outside shifts the graph up by .
- The outside stretches the graph vertically by , and when is negative it also reflects the graph across the x-axis, flipping the curve downward.
Putting these together, the corner that used to sit at now sits at . The domain is , and the range is when or when . The range splits into two cases because the curve leaves the corner heading up or down according to the sign of .
Worked example 2 Graph and state its domain and range
Match the rule to the form . Here , , and , so nothing is stretched or flipped, and the whole curve is only shifted.
The inside moves the graph right by , and the moves it up by , so the corner travels from to
From there the curve has the ordinary square-root shape. To place a second point, pick an input that makes the radicand a perfect square, say : then , giving the point .
The domain comes from the radicand: , so , which matches the corner starting at . Since the curve rises from the corner, so the outputs start at and grow, and the range is .
Worked example 3 Describe the graph of
Read as with , a single coefficient doing two jobs at once.
Its size, , is a vertical stretch: every output of is doubled. Its sign, negative, reflects the graph across the x-axis, turning the curve downward. Tracking the anchor points of through the rule makes both effects concrete:
The corner stays fixed, since . The domain is unchanged at , because the radicand is untouched. But the outputs are now zero or negative, so the range flips to . The curve now drops away from the origin into the fourth quadrant instead of climbing into the first.
Check your understanding
The graph of is the graph of shifted how, and where is its corner?
Match to . Writing as gives , and the constant is .
A negative shifts left by and a negative shifts down by , so the corner lands at .
The three controls below are the three letters of , and the figure restates the domain in words underneath as you move them. The checkpoint you just answered is a state you can reach. Step the inside shift down to and the outside shift down to , and the corner lands where you said it would.
Watch the domain and the range behave differently, because that difference is the point of this section. The inside shift drags the domain with it: the curve simply does not exist to the left of the corner. So moving moves the edge of the world, and the sentence under the figure moves with it. The outside shift does not touch the domain at all, it only raises the whole picture, and the range follows it. Then return both shifts to and set to , which is worked example 3 exactly: the range turns over while the domain sits where it was. Take all the way to and the curve flattens into a horizontal line, and note that it is still a line starting at . It still has nothing to its left: the restriction was never about the shape, it came from inside the root.
The domain and the range of
y = √x. Its start point sits at (0, 0), not shifted at all. From there it rises to the right, at the natural rate, flattening as it goes. It exists only for x of at least 0, because a square root needs the inside to be at least 0.
Solving radical equations
A radical equation has the variable under a root, as in . The plan is to free the variable from the root, and the tool is the inverse relationship you have leaned on all lesson: squaring undoes a square root. Isolate the radical on one side, then raise both sides to the power that matches the index. For a square root you square; for a cube root you cube.
Squaring turned the awkward equation into a linear one. There is a catch, though, and it is not optional bookkeeping. Raising both sides to a power can manufacture solutions that solve the new equation but not the original. These are called extraneous solutions, and the next argument explains exactly where they come from.
Why raising both sides to a power can create extraneous solutions#
Squaring both sides of an equation is a one-way street. If two quantities are equal, then their squares are equal too: from it always follows that . So every genuine solution of the original equation survives the squaring, and none is lost.
The reverse direction is where trouble enters. Knowing does not force , because a square erases a sign. The statement means
since and share the same square. Squaring folds those two possibilities into one, so the squared equation quietly carries the solutions of together with the solutions of . Any candidate that belongs only to the second family satisfies the squared equation while failing the one you started with.
Those intruders are the extraneous solutions. They are not errors in the algebra; they are the price of an irreversible step, because squaring cannot tell from . The only way to know which candidates are real solutions is to substitute each one back into the original equation and keep those that make it true.
Worked example 4 Solve
Isolate the radical before doing anything else. Subtract from both sides so the root stands alone:
Now the two sides are single quantities, so square both to remove the root:
Check the candidate in the original equation, not the squared one:
The check holds, so is the solution. Isolating first mattered: squaring the equation while the was still attached would have left a radical behind.
Worked example 5 Solve and watch an extraneous solution appear
The radical is already isolated, so square both sides. The right side is a binomial, so expand it carefully:
Move everything to one side to get a quadratic, then factor:
so the candidates are and . Squaring may have added a stranger, so test both in the original equation .
For : the left side is and the right side is . The two agree, so is genuine.
For : the left side is and the right side is . Since , the candidate fails, so it is extraneous. It is exactly the intruder the proof predicted: it solves the other branch , because is true. The only solution is .
Worked example 6 Solve
Square both sides, expanding the right side as a binomial:
Collect into a quadratic and factor:
giving candidates and . Test each in the original equation, since squaring might have introduced a false root.
For : , and . They match.
For : , and . They match too.
This time both candidates check out, so the equation has two solutions, and . The lesson is not that the smaller root is always extraneous; it is that you cannot know until you test, so check every candidate in the original equation.
Check your understanding
How many real solutions does have?
Square both sides and expand the right side.
Factoring gives , so the candidates are and . Testing in the original: and agree, but while do not. Only survives, so there is exactly one solution.